Structural analysis of DNA wrapping in bacterial transcription initiation complex by transmission electron microscopy and single particle analysis

Texto

(1)UNIVERSIDADE DE SÃO PAULO INSTITUTO DE FÍSICA DE SÃO CARLOS. ALFREDO JOSE FLOREZ ARIZA. Structural analysis of DNA wrapping in bacterial transcription initiation complex by transmission electron microscopy and single particle analysis. São Carlos 2018.

(2)

(3) ALFREDO JOSE FLOREZ ARIZA. Structural analysis of DNA wrapping in bacterial transcription initiation complex by transmission electron microscopy and single particle analysis. Dissertation presented to the Graduate Program in Physics at the Instituto de Física de São Carlos, Universidade de São Paulo to obtain the degree of Master of Science. Concentration area: Applied Physics Option: Biomolecular Physics Advisor: Dr. Rodrigo Villares Portugal. Corrected Version (Original version available on the Program Unit). São Carlos 2018.

(4) I AUTHORIZE THE REPRODUCTION AND DISSEMINATION OF TOTAL OR PARTIAL COPIES OF THIS DOCUMENT, BY CONVENCIONAL OR ELECTRONIC MEDIA FOR STUDY OR RESEARCH PURPOSE, SINCE IT IS REFERENCED.. Florez Ariza, Alfredo Jose Structural analysis of DNA wrapping in bacterial transcription initiation complex by transmission electron microscopy and single particle analysis / Alfredo Jose Florez Ariza; advisor Rodrigo Villares Portugal - revised version -- São Carlos 2018. 93 p. Dissertation (Master's degree - Graduate Program in Física Aplicada Biomolecular) -- Instituto de Física de São Carlos, Universidade de São Paulo - Brasil , 2018. 1. Transcription initiation complex. 2. RNA polymerase. 3. DNA wrapping. 4. Transmission electron microscopý. 5. Single particle analysis. I. Portugal, Rodrigo Villares, advisor. II. Title..

(5)

(6)

(7) With love to my parents, Norma and Enrique, my brother Diego and my sister Nicole, and to my lovely Annar..

(8)

(9) ACKNOWLEDGMENTS I thank my parents, Norma and Enrique, for all this time of sacrifice. They are a big part of my motivation, and although life has separated us, I am sure that someday we will be able to meet again and be happy. I love them incalculably. To my brothers, Diego and Nicole, who with their joy and their company, made my days happier. They motivate me to be a better person; they encourage me to take on new challenges. I miss them every day. To my love Annar, for these more than five years we are together. It has been complicated to be separated, but your unconditional support, love and great friendship, makes me feel that soon we will be together again. I love you. To my good friend Juan, his brother Sebastian, and to his parents Cesar and Esperanza. Thanks for your constant support offered. All your affection will be engraved in my memory. To my advisor, Rodrigo Portugal, thanks for all the support and trust placed. Although more than one problem occurred, he always motivated me constantly to continue. Without a doubt, his orientation in addition to his excellent human and academic quality, is a reason for reference and admiration for me. To Professor Marin Van Heel, for his willingness and academic advice. The minutes that he uses for us are of great help to improve daily. It is really very rewarding and valuable to be able to count with him in the laboratory. To Professor Daniel Guerra, for all the confidence placed in me during the stage in which I was his student. Undoubtedly, that stage under his academic guidance, allowed me to acquire new knowledge that was vital to get involved in this work. Thank you very much for the opportunity, the support, and the friendship offered. I would like to thank the new friends I met here. To my good friend Marcelo, thank you very much for so much support during these years, for your unconditional friendship and for your good advice. To Alexandre Cassago, for his dedicated support during the beginning of this work, for his dedication and complete willingness to develop this research. To my colleagues and new friends of the LNNano: Antonio, Débora, Daniela, Karo, Juliana, Katye, and Rafael, for the pleasant shared moments and scientific discussions.To the Institute of Physics of Sao Carlos (IFSC), to the National Nanotechnology Laboratory (LNNano) and to the CNPEM for all the infrastructure offered, and to the CAPES agency for its financial support..

(10)

(11) “Science is not only compatible with spirituality; it is a profound source of spirituality.” ―. Carl Sagan, TheDemond-Haunted World: Science as a Candle in the Dark..

(12)

(13) ABSTRACT FLOREZ ARIZA, A. Structural analysis of DNA wrapping in bacterial transcription initiation complex by transmission electron microscopy and single particle analysis. 2018. 93p. Dissertação (Mestrado em Ciências) Instituto de Física de São Carlos, Universidade de São Paulo, São Carlos, 2018. The transcription initiation is the first step in gene expression and an important regulation step in all living organisms. In bacteria, it has been proposed that DNA bending and its wrapping on the surface of E. coli RNAP might facilitate the opening of the transcription bubble, which is necessary for the initiation of gene transcription. In this work, it is shown the first structural study to evaluate a DNA wrapping model, including its length and the relative position in the bacterial transcription initiation complex (RP complex), assembled between RNA polymerase-σ70 holoenzyme (RNAP) and a λPR promoter (-100 to +30 wild type). RP complex was prepared and negatively stained with 2% uranyl acetate on a thin-carbon coated grid and the data acquisition of 500 images was performed in a JEM-2100 (JEOL, Japan) microscope equipped with an F416 CMOS camera (TVIPS, Germany). Single particle analysis of 16,015 particles, grouped in 666 class-averages, was conducted using IMAGIC 4D software (Image Science, Germany) to obtain a three-dimensional model of the RP complex at 20Å resolution. After the rigid-body fitting of the RNAP crystallographic structure (PDB 4YG2) and the modeled DNA promoter, it was observed that the regions 1.2 and 4.2 of the σ 70 subunit interacts with the consensus zones, -10 and -35 hexamers of the promoter. Furthermore, it was possible to observe that αCTDs (C-terminal domain) in both alpha subunits would be oriented to facilitate the interaction with the first and second UPelements regions, respectively (centered around –50 and -75 positions in the promoter). These was enabled by the presence of the characteristics motifs helix-hairpin-helix in these domains. In addition, the downstream DNA, from the transcription bubble, appears to be inside the protein main channel, oriented in a way to enable interactions with the RNAP clamp and jaws. Finally, it was observed that the DNA wrapping has ~32 nm of total length and involves a promoter bent of ~255° around the RNAP surface. The 3D-model obtained in this study is the very first direct structural confirmation of the DNA promoter wrapping in a bacterial transcription initiation complex. Keywords: Transcription initiation Complex. RNA polymerase. DNA wrapping. Transmission electron Microscopy. Single particle analysis..

(14)

(15) RESUMO FLOREZ ARIZA, A. Análise estrutural do enovelamento do DNA no complexo da iniciação de transcrição bacteriano usando microscopia eletrônica de transmissão e análise de partículas isoladas. 2018. 93p. Dissertação (Mestrado em Ciências) - Instituto de Física de São Carlos, Universidade de São Paulo, São Carlos, 2018. A iniciação da transcrição é o primeiro passo na expressão gênica e importante ponto de regulação em todos os organismos vivos. Em bactérias, foi proposto que o enovelamento do DNA na superfície da RNAP de E. coli pode facilitar a abertura da bolha de transcrição, necessária para o início da transcrição gênica. Neste trabalho, é apresentado o primeiro estudo estrutural direto para avaliar o comprimento do enovelamento do DNA e sua posição no complexo de iniciação da transcrição bacteriana (complexo RP), montado entre a holoenzima RNA polimerase-σ70 (RNAP) e um promotor λPR (-100 para +30, tipo selvagem). Amostras do complexo RP foram preparadas e contrastadas negativamente com 2% de acetato de uranila em uma grade com filme fino de carbono e a aquisição de 500 imagens foi realizada em um microscópio JEM-2100 (Jeol, Japão) equipado com uma câmera CMOS F416 (TVIPS, Alemanha). A análise de partículas isoladas de 16.015 partículas, agrupadas em 666 médias de classe, foi conduzida usando o software IMAGIC 4D (Image Science, Alemanha) para obter um modelo tridimensional do complexo RP, a 20Å de resolução, estimado pelo critério de ½ bit. Após o ajuste de corpo rígido da estrutura cristalográfica da RNAP (PDB 4YG2) e do promotor de DNA modelado, observou-se que as regiões 1.2 e 4.2 da subunidade σ 70 interagem com as zonas de consenso, hexâmeros -10 e -35, do promotor. Além disso, foi possível observar que os αCTDs (domínio C-terminal) em ambas as subunidades alfa estariam orientados para facilitar uma possível interação com a primeira e segundas regiões dos elementos UP, respectivamente (centradas em torno das posições –50 e -75 do promotor). Estas seriam possíveis devido à presença de alguns motivos de características hélice-grampo-hélice nesses domínios. Além disso, a região do promotor, downstream da bolha de transcrição, parece estar dentro do canal principal da proteína, orientado de forma a possibilitar interações com o clamp e jaw da RNAP. Finalmente, foi observado que o comprimento total do enovelamento de DNA envolve cerca de 32 nm e 255° de rotação do DNA ao redor da superfície da RNAP. Portanto, este modelo 3D é a primeira confirmação estrutural direta do enovelamento de DNA em um complexo bacteriano de iniciação da transcrição. Palavras-chave: Complexo da iniciação da transcrição. RNA polimerase. Enovelamento do DNA. Microscopia eletrônica de transmissão. Análise de partículas isoladas..

(16)

(17) LIST OF ABBREVIATIONS AND SYMBOLS. CCD. Charge-coupled Device. CTF. Contrast Transfer Function. TEM. Transmission electron microscopy. SPA. Single Particle Analysis. MRA. Multi-reference Alignment. MAS. Multivariate Statistical Analysis. ABC. Alignment by Classification. PDB. Protein Data Bank. αCTD. alpha subunit Carboxi-terminal domain. DNA. Deoxyribonucleic acid. RNAP. RNA polymerase. RP. Transcription inititation complex.

(18)

(19) CONTENTS 1. INTRODUCTION ........................................................................................................................ 17. 1.1. Bacterial transcription initiation complex ......................................................................... 17. 1.1.1. General aspects .......................................................................................................................... 17. 1.1.2. The σ70 factor ................................................................................................................................ 18. 1.1.3. Alpha subunits and UP-elements........................................................................................ 20. 1.1.4. The "wrapping" phenomenon................................................................................................ 22. 1.2. Technique: Transmission Electron Microscopy............................................................ 25. 1.2.1. Overview ......................................................................................................................................... 25. 1.2.2. Sample Preparation ................................................................................................................... 29. 1.2.2.1 Negative Stain .............................................................................................................................. 29 1.2.2.2 Cryogenic sample preparation ............................................................................................. 29 1.2.3. Instrumentation ............................................................................................................................ 30. 1.2.3.1 Electron gun .................................................................................................................................. 32 1.2.3.2 Wehnelt cylinder .......................................................................................................................... 33 1.2.3.3 Condensor lens............................................................................................................................ 34 1.2.3.4 Objective lens ............................................................................................................................... 35 1.2.3.5 Intermediate and projector lens ........................................................................................... 35 1.2.3.6 Fluorescent screen and detectors ...................................................................................... 35 2. OBJECTIVES ............................................................................................................................... 37. 2.1. General objective ........................................................................................................................ 37. 2.2. Specific objectives ...................................................................................................................... 37. 3. EXPERIMENTAL PROCEDURE AND METHODOLOGY ...................................... 39. 3.1. Image Processing ....................................................................................................................... 39. 3.1.1. Preparing the micrographs ..................................................................................................... 41. 3.1.2. CTF correction ............................................................................................................................. 42.

(20) 3.1.3 Particle picking............................................................................................................44 3.1.4 Multivariate statistical analysis (MSA): Hyperspace data compression and Classification………………………………………………………………………...45 3.1.5 Euler angular assignment.............................................................................. ...47 3.1.6 Three-dimensional reconstruction....................................................................49 3.2. DNA production................................................................................................50. 3.3. Transcription Initiation complex assembly……………………… ……………..…….51. 3.4. Specimen preparation and TEM imaging………………………………………...52. 3.5. Single particle analysis……………………………………………………………..53. 3.6. Modelling………..…………………………………………………………………...54. 4. RESULTS AND DISCUSSIONS…………………………………………………..55. 4.1. Transcription Initiation Complex Assembly and TEM imaging…………………57. 4.2. Image processing – Single particle analysis……………………………………..56. 4.2.1 Micrographs preparation and Contrast Transfer Function correction …………....56 4.2.2 Particle picking……………………………………………………………………….59 4.2.3 Multivariate Statistical Analysis and classification…………………………………60 4.2.4 Angular reconstitution………………………………………………………………62 4.2.5 Three-dimensional reconstruction…………………………………….………………63 4.2.6 Iterative refinement loop……………………………………………………………65 4.2.7 Final 3D reconstruction…………………………………………………………………..69 4.3. Model fitting…………………….……………………………………………….………….70. 4.3.1 RNA Polymerase rigid body fitting and DNA modeling…………………….……….70 4.3.2 Structure of the Transcription Initiation complex with wrapped promoter… …..71 4.4. Promoter wrapping formation and RPo assembly………………………………….75. 4.5. DNA wrapping: agreement of the EM structure with other proposed models ...79. 5. CONCLUSIONS AND FUTURE PERSPECTIVES……………………………...…...85 REFERENCES……………………………………………………………………….……87.

(21) 17. 1. INTRODUCTION. 1.1. Bacterial transcription initiation complex. 1.1.1 General aspects. The assembly of transcription initiation complex (RP) is the first step for the gene expression and an important regulation point for all living organisms. One of the major players in the formation of this complex is the RNA polymerase holoenzyme (RNAP), which is formed from the binding of the core enzyme with the sigma factor. Bacterial core RNA polymerase consists of five subunits, α2ββ′ω, with a molecular mass of 370 kDa (Figure 1).1 This core enzyme is capable of nonspecific DNA binding and starts the RNA synthesis from DNA ends or nicks. The transcription machinery must recognize with high specificity the sites where genes begin, called promoters, among a huge extension of non-specific binding sites offered by genomic DNA. The core enzyme needs to bind the sigma factor (σ) to form the transcriptionally active holoenzyme. This σ factor has the capability to specificity recognize some consensus regions on the DNA promoter and thus start the process of forming a stable RP complex.2,3 The nature of the sigma factor with which the core enzyme would interact depends on the environmental conditions and the identity of the promoter. Bacteria possess an essential housekeeping σ factor that controls many different promoters. In Escherichia coli it is the σ70 (molecular weight of 70 KDa) and in Bacillus subtilis and other Gram-positive bacteria it is the σA.4 In the case of the σ70 factor, it recognizes the -35 and -10 hexamers consensus zones, upstream from the transcriptional start site (+1) of the promoter (Figure 1).5 The sigma factor binding to the consensus zone results in the formation of the closed complex (RPC).That induces significant conformational changes in both, the RNAP and the DNA, including the transcription bubble formation, to form a stable transcription initiation open complex (RPo).2,5.

(22) 18. Figure 1 - Crystallographic structure of E. coli RNA polymerase sigma 70 holoenzyme (left: PDB 4YG2). The σ70 is in red, ω in yellow, β in magenta, β’ in green, and the alpha subunits are shown as follow: αI in blue and αII in light-blue. Structure of the close RP complex between the RNAP holoenzyme and fork-junction DNA. The core enzyme is shown in surface representation and the sigma factor is shown as a Cα backbone worm. It can be observed the -35 and the -10 elements (in yellow) are recognized by the region 4 and 2 of the sigma subunit respectively. Source: Adapted from MURAKAMI;6 MURAKAMI; DARST.2. All these conformational changes are associated with kinetic intermediates which vary in stability and lifetime among different promoters, suggesting that the DNA sequences determine different configurations of the RP complex. 7. 1.1.2. The σ70 factor. The sigma factor plays a crucial role in the specific promoter recognition. The σ70 subunit has four different domains or regions (from 1 to 4) and one nonconserved region (NCR). Each of these regions have different sub-domains (Figure 2). There are several biochemical and structural studies supporting the idea that the DNA recognition and binding, is accomplished by the interaction of the σ4.2 subdomain with the -35 hexamer, while σ3.0 sub-domain binds to the -10 hexamer. The σ1.2 and σ2.1 – σ2.4 regions interact with the transcription bubble8 and the σNCR interacts with the promoter DNA upstream of the -10 element.9.

(23) 19. The TGn element on the promoter, a consensus region which has been reported be crucial for an optimal activity for different promoters is recognized by the σ3.0 subdomain, possibly through E458 and H455 residues as was shown by some mutagenesis assays.10 This interaction was also shown to be present in T. aquaticus RNAP holoenzyme-fork DNA complex, through analogous residues in the σA factor.11. Figure 2 - (A) Sketch of the E. coli σ70 sub-domains. (B) Schematic representation of the DNA canonical promoter regions recognized by the σ70 sub-domains to form the RP complex. Source: Adapted from FEKLÍSTOV;5 MURAKAMI; DARST.2. Furthermore, the σ1.1 sub-domain seems to perform a regulatory role. Studies using NMR spectroscopy showed that in absence of RNA core enzyme, the σ70 is auto-inhibited to DNA binding. It is suggested that there is an "intra molecular interaction" between the 4.2 sub-domains, which should bind to -35 hexamer, with the region σ1.1.12 Moreover, there is the hypothesis that when the σ70 binds to the DNA, the σ1.1 blocks the “primary” or main channel of the enzyme, during the formation of the transcription initiation complex. After major conformational changes, the σ1.1 is expelled, allowing the DNA promoter to completely enter the primary channel (Figure 3). This hypothesis is supported by the structural study of the Phage T7 Gp2 protein inhibition of the RNAP-σ70, which shows that there is a ternary interaction between the σ1.1-Gp2-β'jaw, that prevent the σ1.1 to leave the main channel inhibiting the function of the enzyme.13.

(24) 20. Figure 3 - (A) Cartoon representation of conformational changes in the transition from RPc to the RPo complex. In the figure the core E. coli RNA polymerase is shown in grey – except the β-flap shown in blue- and the σ70 factor is in orange. Each σ domain is indicated by name. Once the RPo complex is formed, the σ1.1 is out of the main channel. Source: Adapted from MURAKAMI; DARST. 2. Some of the conformational changes, occurred during the formation of the transcription initiation complex, involve portions of DNA that are considerably distant to the transcription-starting site. These regions on the promoter, known as UP-elements, are upstream regions to the -35 hexamer, and generally possess phased A-T tracks with a certain length that could interact with the α subunits of RNAP.. 1.1.3 Alpha subunits and UP-elements. The RNAP has two alpha subunits both have the same polypeptide chain. Each α subunit has two independent folded domains: the amino-terminal domain (αNTD, residues 8–233), responsible for dimerization and interactions with β and β’, and the carboxyl-terminal domain (αCTD, residues 245–329). A flexible tether, known as α-linker, of about 18 residues length, connects the two domains.14 There is experimental evidence that both αCTDs have a key role in the formation of transcription initiation complex. These are capable of binding DNA specific sequences, commonly called UP-elements. These UP-elements, located upstream the -35 hexamer, are not conserved as high as the -10 and -35 hexamers, but exhibit interesting features, as to own A-T rich regions (of ~10 bp length or more,.

(25) 21. separated with a periodicity of ~10.3 bp), providing a particular curvature to the DNA, which seems to be related with the promoter strength and transcription regulation.15,16 In this sense, biochemical and genetic screening studies showed that the αCTDs can bind UP-elements mainly through seven residues: L262, R265, N268, C269, G296, K298, S299, all of which belong to a helix-harping-helix motif present in this domain.17 Even more, X-ray crystallography studies showed that αCTD is oriented to interact with the DNA minor-groove and backbone of the UP-elements. Moreover, because these UP- elements can be in close proximity to the -35 hexamers, the αCTD could be possibly interacting with the region 4 of the σ factor, aiding to stabilize the RP complex.18,19. Figure 4 – Crystallographic structure of the ternary complex CAP-DNA-αCTD. The CAP protein is in green, the αCTD in yellow and the DNA UP-element region is in cyan. In the figure can be observed that the residues (in spheres) lys-298 (red), arg-265 (magenta) and asn-268 (grey), are oriented to interact with the minor groove and backbone of the DNA UP-element. Source: Adapted from BENOFF.18.

(26) 22. The effects of the UP-elements on promoter strength were established by different studies that evaluated how different upstream sequences can affect the relative activity among promoters. In this sense, Ross and co-workers showed that the two UP- elements (proximal UP-element, -38 to -41 and distal UP-element, -51 to -60) of the wild-type rrnB P1 promoter, improved the in vivo transcription activity of the lacZ core promoter (+52 to -37) up to ~33 fold when having both UP-elements.20 But even each one of the UP- elements, separately, were able to improve the transcription activity. Moreover, it was reported in the same study that E. coli RNAP holoenzyme lacking αCTDs was not able to interact with the UP-element regions (confirmed by foot-printing assays), and that the transcription activity was completely dependent on this interaction.20 Furthermore, this kind of UP-elements/αCTDs interactions seems to be present in different bacterial phyla. For example, a study performed in Helicobacter pylori, by Nuclear Magnetic Resonance (NMR), showed that both αCTDs interact directly with the UP-element of the DNA promoter and NikR protein to form a stable ternary transcription initiation complex.21. The interactions described above, involve DNA regions centered on -45 to 60 bp upstream the transcription start site. However, for some promoters, like the λ PR promoter, it was reported that the RNAP-αCTDs was able to interact with far upstream zones of the DNA, up to -90 bp upstream. This kind of new interaction between those UP- elements and RNAP would only be possible if the DNA promoter underwent a significant curvature around the RNAP surface. This interesting and relatively new phenomenon, known as DNA wrapping, will be detailed in the following section.. 1.1.4. The "wrapping" phenomenon. As mentioned before, the interaction between those upstream promoter regions and RNAP would only be possible if the DNA promoter underwent a significant curvature around the RNAP surface, hinting a DNA wrapping formation. In this sense, experimental evidence suggest that DNA wrapping would entail a DNA curvature of ~300° around the RNAP surface (Figure 5).22 Concerning the λPR promoter, different studies by Atomic Force Microscopy (AFM), proposed that in the RP complex, the promoter was forming a "wrapping" of ~30 nm (~90 bp).

(27) 23. around the RNAP surface (Figure 6). Also, it was reported that this DNA wrapping is reduced to ~ 3 nm when the complexes are formed with RNAP lacking the αCTDs, and to ~15nm when the α-linker is missed. Those observations seemed to be according with different DNase and OH radical foot-printing assays, performed in previous studies. 23–26. Figure 5 -. A square patch of one AFM micrograph of RP complexes (top left) assembled between E. coli RNAP and a DNA fragment containing the λ PR wild type promoter (-100 to +34). A zoom in image (top right) showing the RP complexes and the measures of the DNA contour length free and in complex. Schematic representation of the RP complex (bottom), showing the RNAP in light grey and the DNA promoter in dark grey. The DNA trajectory has proposed around the RNAP surface, with a bend angle of 300°. Source: Adapted from RIVETTI.22. Recently, a study using fluorescence resonance energy transfer (FRET), on the same system, measured the distance between fluorescent labels in positions +14 and -100 relatives to the transcription starting site (+1), confirming that these locations come close to each other in a way that is consistent with the proposed DNA wrapping.27.

(28) 24. Figure 6 – (A) Contour length distributions of the DNA both in absence (dashed bars, top) and presence of the RNAP (gray bars, bottom). The DNA compaction in the RP complex is around 30 nm. (B) Cartoon representation of the RP complex, showing that UPelements, centered around -50 and-90 bp, would be interacting with the αCTDs Source: Adapted from MANGIAROTTI.23. Furthermore, another study showed that the DNA wrapping length is significantly reduced by allosteric regulators like the guanosine tetra phosphate (ppGpp) and Dksa protein.28 The wrapping phenomenon have been also suggested to be present in other promoters like rrnB P1, hdeAB and tRNAtyr;29–31 and also for eukaryotes, as was observed for the adenovirus promoter in complex with the RNA pol II of S. cerevisiae.32,33 All of these studies, seem to indicate that the DNA wrapping is a conserved phenomenon along different kind of promoters and directly related to the promoter strength and stability of the transcription initiation complex. However, even with all the studies mentioned above, there is no direct structural evidence of the DNA wrapping. Until now, there are questions that remains without answer like which could be the protein domains possibly interacting with all the UPelements? How could this be related to the formation of a stable RP complex? Which would be the real extension of the DNA wrapping, and the promoter regions involved? In the present work, we present the first direct structural study of DNA wrapping in the RP complex assembled between the E. coli RNAP holoenzyme (RNAP-σ70) and the λ PR wild- type promoter (-100 to +30 bp). This study was carried out by transmission electron microscopy (TEM) and single particle analysis (SPA)..

(29) 25. 1.2. Technique: Transmission Electron Microscopy. 1.2.1. Overview. The relationship between structure and function is evident at all levels of the life sciences. Just as evolution has spawned fins, legs, wings, and hands as divergent variations of the same pattern, the bio-machines that make the cell work, are each one made up of molecules with specific structures for their purposes. The enzymes receive the substrate and accommodates it oriented to the functional groups that ensure that the desired chemical reaction takes place. Cytoskeletal proteins are formed as large girders that can assemble networks with the rigidity or flexibility necessary to support the structure and movement of the cell. Molecular motors couple chemical reactions and the use of thermal energy to move in a specific direction to, for example, transport vesicles between organelles. Structural biology proposes to elucidate these and many other processes by solving the functional structure of these macromolecules of life, with atomic resolution. Since the 1950s, X-ray diffraction has been the predominant technique to carry out this task. Providing a crystal formed by periodic and well-organized repetitions of the molecule under study, it is possible to elucidate its 3D structure with atomic resolution. However, many of these molecules may not form crystals. There are also some limitations to elucidate flexible domains of the molecule under study.34 On the other hand, nuclear magnetic resonance (NMR) is based on the detection of the atoms chemical displacement, and therefore their bounds to each other, allowing for the 3D reconstruction. However, when molecules like proteins, are larger than 35-50 kDa, the chemical shift signals may overlap each other, making it challenging or impossible to differentiate them.35 Currently, Single Particle Cryogenic Electron Microscopy (Cryo-EM) is capable to achieve resolutions comparable to X-ray crystallography, with the advantage of not requiring the crystallization of the sample and the capacity, in some cases, of giving direct information about different conformational states associated to the biological function of a molecular complex..

(30) 26. Figure 7 - Schematic representation of how Cryo-EM works. An electron beam, accelerated with 100-300 kV, interacts with frozen protein sample, and generate 2D projections that shown the protein in different random orientations, which will use then to generate a 3D model.. Source: Adapted from CRESSEY; CALLAWAY.36. The development of Cryo-EM technique started in the 1980s,37,38 and for many decades it was referred as a "blobology" science, since it was only able to produce low-resolution maps (15-20 Å), unlike the high-resolution achieved by Xray crystallography. However, the continuous improvements in methods and instrumentation led to a key situation in 2013 (Figure 8). It was then that the socalled "resolution revolution" was initiated, driven by the new generation of direct electron detectors that present a very high speed and sensitivity, superior to previously used CCD devices.39–41 With a better signal-to-noise ratio and the ability to produce films instead of single shots, in recent years, Cryo-EM has reached the resolution of traditional X-ray crystallography.42,43. Figure 8 - Improvement of the resolution achieved by Cryo-EM in last few years. From ~20Å before the 2013 year until better than 2Å up today. Source: Adapted from CRESSEY; CALLAWAY.36.

(31) 27. In Figure 9, it is shown a graphic summary of a recently reported structure of the hisPEC-NusA complex, obtained by Single particle Cryo-EM with an overall resolution of 3.6 Å. 44. Figure 9 - A micrograph patch of the hisPEC-NusA complex obtained by Cryo-EM (A). Some representative class- averages, showing the complex in different orientations (B). A 3D model reconstructed at ~3.6 Å of global resolution in which is clearly possible observe the amino-acids backbone and nucleic acids electron density (C). The scale bar is 100 nm. Source: Adapted from GUO. 44. During the last 15 years, thousands of structures have been produced by CryoEM. Figure 10 shows the number of density maps for different macromolecular complexes, deposited per year (from 2002 to date) in the public repository “The Electron Microscopy Data Bank” (EMDB).45 In addition, the resolution achieved for the density maps has been improving year after year, as shown in Figure 11. The improvement trend can be observed in the curves of highest resolution and average obtained during each year (from 2002 to today), for the different density maps deposited in the EMDB. 45.

(32) 28. Figure 10 - The cumulative number of 3D maps released (2002 to 2018) Source: Adapted from EMDB STATISTICS.45. Figure 11 - The resolution trends of Single particle Cyo-EM released maps from 2002 up to April 2018. The red line indicates the average resolution attained in each year. In blue line is shown the highest resolution attained in each year. In the small panel it is shown the segmented surface map of the Eukaryotic Translation initiation factor 2B (~80 kDa), which the structure was solved at 2.8 Å of resolution in April 2018. Source: Adapted from EMDB STATISTICS.45.

(33) 29. 1.2.2 Sample Preparation 1.2.2.1 Negative Stain. The negative stain method was introduced by Brenner and Horne in 1959, while they were studying the external structure of viral particles.46 Since the beginning, this technique was widely used to obtain high contrast imaging of different biological complexes. The method consists in mixing a heavy metal salt solution, usually uranyl acetate or uranyl molybdate at 1% or 2% (m/v), with the solution containing the sample. This method provides a high contrast image, but with some downsides. For example, the sample is exposed to the hash vacuum environment of the microscope, together with the large size of the salt crystals and even the access of the staining agent to hydration channels, the structural information that can be obtained is limited to a low resolution. Nowadays, the negative stain is still used to carry out the screening of the sample (2D) and 3D analysis of complexes when the necessary information can be obtained at low resolution. As a screening tool, it is useful for assessing the sample homogeneity, purity and overall quality of the protein or biological complex. 47 Furthermore, if a single particle analysis is performed of these images, it is possible to obtain a 3Dmodel reconstruction at ~20 Å of resolution.. 1.2.2.2 Cryogenic sample preparation. The sample preparation for Cryo-EM consists, in summary, in applying the sample solution onto the grid, removing the excess of liquid and rapidly plunging the specimen into a vessel with liquid ethane, previously thermalized by a liquid nitrogen bath. Due to the fast freezing process, the water molecules of the buffer, will not form a crystalline structure, hexagonal or cubic, forming an amorphous ice, or vitreous ice. This vitreous ice will preserve the protein in its native state and at the same time provide a partial protection against electron beam damage. In this way, the molecules under study will be "embedded" in a layer of amorphous ice, distributed in different random orientations (Figure 12). The subsequent imaging stage will produce several 2D-projections of the particle, allowing to record enough data to achieve a 3D.

(34) 30. reconstruction of the biomolecule or complex under study. Since the first reports of vitreous ice specimen preparation, Cryo-EM has undergone a "quantum leap" progression in its attainable resolution and applicability to the study of challenging biological systems.37,38,48–50. Figure 12 - Schematic representation of samples vitrification for Cryo-EM. This method was introduced by Jacques Dobuchet around 1980’s. Source: Adapted from NOBELPRIZE-ORG.51. 1.2.3. Instrumentation. The basis of transmission electron microscopy (TEM) is the use of a high-energy electron beam (typically accelerated between 100-300 kV) passing through the sample in a way to gather information of its intern structures. Because of the particle-wave nature of electrons, these can interact with matter in different ways. The image formation arises from a combination of the interaction of elastic-scattered.

(35) 31. electron-waves, non-scattered electron-waves and inelastic scattered electronwaves. It is important to mention that the wavelength of an electron accelerated at 300 kV -like in TEM- is around 2 pm.52 Therefore, the attainable resolution, typically around a few angstroms, is not limited by the wavelength of the radiation. The resolution limitation is a result of many different aspects, but mainly the low electron doses used during the image acquisition. This use of low electron-doses causes the shot-noise, or quantum-noise, to be the main reason of having images with a signal/noise ratio much smaller than the unit, limiting in this way, the possible resolution to reach. Obviously, through modern methods of images analysis as well as the recent developments of instrumental improvements, like the direct electron-detectors, these limitations can be partially overcome.53,54 The typical transmission electron microscope is generally composed by an electron gun, a set of condenser and objective lenses, intermediate lenses, projector lenses, a fluorescent screen on the bottom and detectors. In the figure 13, it is shown an overview diagram of the internal structure of a transmission electron microscope. Some of the components will be discussed later..

(36) 32. Figure 13 – An image of an Electronic transmission microscope (left) and a generalized design showing its internal structure (right). Source: Adapted from AMMRF.55. 1.2.3.1. Electron gun. The electron gun or cathode, function as an electron beam emitter, seating up on the top of the microscope column. It is kept at the acceleration voltage, for instance 300 kV, while the rest of the microscope is kept at ground potential. Since the electrons are emitted in a divergent way, the cathode filament is surrounded by a cupshaped electrode, with a central of approximate 1 mm bore, called Wehnelt cylinder, which has the function to conduct the electron beam emitted from the cathode toward the optic axis. It is the tip of this cloud, closest to the Wehnelt cylinder opening that acts as the actual electron source. The electron guns are classified in two types, the thermionic sources (tungsten filament and lanthanum hexaboride, LaB6) and the field emission guns (FEG). The principal difference between the sources are the brightness and energy spread (Figure 14), which are both directly.

(37) 33. related to the resolution that can be reached.56. Figure 14 -. Top: Schematic representation of the electron gun, with diverse types of emitters (source), tungsten filament (a), LaB6 (b) and FEG (c). Bottom: a table showing the principal differences between the three electron sources. Source: Adapted from AMMRF.55. 1.2.3.2. Wehnelt cylinder. To drive the electrons emitted by the gun toward the optical axis, the cathode filament is surrounded by a cup-shaped electrode that is kept at a slightly more negative potential that the electron source (~500V less). This electrode, called Wehnelt cylinder, has a small ~ 1 mm central bore, through which the electrons will be ejected.56.

(38) 34. Figure 15 - The Wehnelt cylinder after to be removed from an electron gun Source: Adapted from AMMRF.55. 1.2.3.3. Condenser lens. The condenser lenses purpose is to manage the electron beam towards the desired direction. Since the electrons are charged particles, the lenses for an electron microscope are electromagnetic lenses, made of ferromagnetic materials, having the capacity of generating electromagnetic fields by a circulating internal current. The generated fields bend the electron beam path to the optical axis. The condenser lenses, located immediately below the Wehnelt, have the objective to control the electron beam that is coming out from the electron gun, thus achieving a better coherence in the illumination on the sample.. Figure 16 - Electromagnetic lenses from an electron microscope (left) and a cartoon showing how could be the electron beam path due to the electromagnetic field generated byb the lenses Source: Adapted from AMMRF.55.

(39) 35. 1.2.3.4 Objective lens This set of lenses is the first that participate directly in the image formation process. Due to its small focal length (only some millimeters), the sample to be imaging is immersed inside the bore of this lens. It is important to mention that in the back focal plane of this lens, the scattered parallel beams are focused in a single point, thus, a "diffraction pattern" of the sample is formed on this plane. Using another set of lenses, the diffraction information can be recombined to form the image of the specimen. In addition, the objective aperture -a physical aperturethat is positioned just below this lens, aid to eliminate the widely scattered beams, forming what is called amplitude contrast. Moreover, the electron beam suffers a phase shift while passing through the sample, also known as phase contrast. This phase contrast is what allows, in practical terms, to visualize the particles of the biomolecules under study.56–59 1.2.3.5 Intermediate and projector lens The set of intermediate lenses are capable of impose different magnifications while varying its internal current. Usually, for screening and data collection, the magnification range is about a few hundreds to tens of thousands, respectively. It changes according to the experimental conditions, microscope and position of the camera. As their name suggests, these set of lenses, project the final image into the screen or the detector.56 1.2.3.6 Fluorescent screen and detectors A fluorescent screen is a metal plate coated with a thin (range 10–20 mm, depending on voltage) layer of a fluorescent material that converts impinging electrons into bursts of light. It is placed into the image plane and serves as a tool for optimizing the operation conditions and screening before starting with data collection. Devices like CCD or CMOS cameras and direct electron detectors process the signals from the electron beam that had interacted with the sample, allowing the digital image recording. It is important to mention that unlike shown in Figure 13, for Single Particle Cryo-EM experiments, the cameras are typically.

(40) 36. mounted bellow the fluorescent screen. 56,60.

(41) 37. 2. OBJECTIVES. 2.1. General objective. The objective of this work is to structurally study the DNA wrapping phenomenon in the transcription initiation complex (RP) to better understand the interactions between the RNAP and the DNA, suggesting a possible conformation of the wrapped DNA. The RP will be assembled between the E. coli RNA polymerase-σ70 holoenzyme and the λ PR wild-type (-100 to +30) promoter, and the structure will be obtained by transmission electron microscopy (TEM) and single particle analysis (SPA),. 2.2 ➢. Specific objectives. Amplify the 186 bp length DNA fragment containing the lambda PR. wild-type promoter by polymerase chain reaction (PCR). Optimize the RP complex assembly conditions and the TEM grids preparation using negative stain method; ➢. Perform the data collection and image processing using the SPA to. obtain a 3D map of the RP complex; ➢. Obtain a model of the RP based on molecular fitting of the available. RNAP crystallographic structure and an in silico modeled DNA promoter, into the 3D density map previously obtained by the SPA; ➢. Analyze the orientation of different protein domains that could be. interacting with the DNA promoter enabling the wrapping formation; ➢. Investigate and determine the measure of the wrapping length, as well. as the possible bent angle of the promoter, and compare the results with previous reported measures..

(42) 38.

(43) 39. 3. EXPERIMENTAL PROCEDURE AND METHODOLOGY. 3.1 Image Processing In Single Particle Analysis (SPA), the 3D map of the macromolecule is a result of many iterative rounds of alignment, classification, average and angular assignment of several images of particles in random orientations and possibly in multiple conformations. To carry out this task, different software packages are available: IMAGIC 4D, RELION, EMAN, SPIDER, etc.,61–63 each one perform the image analysis with different philosophies. In this section, the different steps in image analysis will be discussed, focused on methods implemented in IMAGIC 4D software (Image Science, Germany), used in the “Align by Classification” (ABC) processing strategy.48,64 Before starting the image processing description, it is useful to mention some important general concepts and examples of practical applications about the Fourier transform, since it is widely used through the image processing. The Fourier transform is a mathematically complex operation that decomposes a function, in real space, into component frequencies represented in the reciprocal space. For an integral and continuous function f: R → C, the Fourier transform will be defined as in Equation (1):. So, F(K) is the Fourier transform of f(x), where K is a real number that express the frequency and х is the variable of the function f in the real space. In addition, i is the imaginary unit. As it was mentioned before, it is denoted that the function f is represented in the real space and its Fourier transform (F) is depicted in the reciprocal space. Therefore, the Fourier transform can be represented as a complex number: F = x + yi or F=|F|(cosθ + i sinθ), where “x” and “y” are the real and imaginary part of the Fourier transform, respectively, and the angle θ is known as the complex argument or phase. In the images below (Figure 17), there are some practical examples about the Fourier transform for 2D-images and how the low and high spatial frequencies are related with different features of the image in the real space..

(44) 40. a. b. c. Figure 17 - (a) The 2D image of the Alan Turing’s protrait (real space) and its 2D Fourier transform (2DFT) in the reciprocal space. The ƒ indicates the application of the forward 2D Fourier transform. The same portrait after to apply a low-pass filter (b) and its corresponding 2DFT. In (c) it is shown the Touring’s portrait after to apply a high-pass filter, and its corresponding 2DFT.. Source: By the author. In Figure 17a, it is shown an Alan Turing’s portrait (left) and its respective Fourier transform (right). Considering the Fourier transform of the image, it is important to note that the low spatial frequencies are represented by those intensities closer to the zeroorder frequency, which is the center of the reciprocal space. These frequencies preserve the information about the shape, size or in other words, the “larger” features of the image. On the other hand, the high spatial frequencies are represented by the intensity spots in the outer areas of the Fourier transform, and these are related to the “finer” or “smaller” details of the image. In Figure 17b, it is represented the original portrait. after. applying. a. low-pass. filter,. and. its.

(45) 41. corresponding Fourier transform (right). As it can be observed, the image only preserves its coarse characteristics. Figure17c shows an example of a high-pass filter in which the low spatial frequencies, closer to the center in the Fourier transform, were suppressed. As a result, it is difficult to observe in the recovered image the borders and limits, but it is conserved the fine details of the portrait. The image processing will be detailed systematically in the next sections.. 3.1.1 Preparing the micrographs. After collecting the micrographs and converting to the appropriate format, it is necessary to prepare these images before starting the analysis. Usually, the very “large” details (i.e., density ramps) and the extremely “small” details (mostly noise) are hiding the motif -the particles- which the analysis is interesting in. The preparation of the micrographs consists in applying a filter to suppress, at the same time, these extreme frequencies in the reciprocal space, which are associated with unwanted information of the image. This operation is performed through a bandpass filter. There are some aspects that should be considered when a band-pass filter is applied. The values of the low-pass (LP) and high-pass (HP) filters shall be chosen according to the particle size as well as the resolution that is expected to be achieved in the analysis. Also, the inversion of densities may be necessary depending on how the software understands the white and black levels as signal. In the case of IMAGIC 4D, it is necessary to have white particles on the gray background. For a common camera setup, if the micrographs are a Cryo-EM dataset, an inversion of densities should be performed. In the same conditions, if the micrographs came from negatively staining samples, invert densities will be not necessary..

(46) 42. Figure 18 -. A square patch of a negatively stained sample, of a protein embedded in a lipid membrane. The figure shows the micrograph before (left) and after (right) applying the band-pass filter. The pixel size is 2.58 Å Source: By the author.. In Figure 18, it is shown a micrograph before (left) and after applying the corresponding band-pass filter (right). The filtered micrograph shows how the background information is reduced as well as the noise. 3.1.2. CTF correction. The contrast transfer function (CTF) is a mathematical representation that explains how the total phase shift of the electron wave, after interacting with the object, is translated into amplitude modulation in the image. For some spatial frequencies, the phase-shift will be “translated” as positive or negative amplitude modulation in the CTF. So, it is necessary to “recover” the information present in the negative spatial frequencies, inverting the amplitudes. Otherwise, the information at these spatial frequencies will contribute to the image formation in a “negative” way. The correction is performed by flipping the phases of the negative lobe of the CTF.48,58,59 The process begins with the calculation of the Fourier transform (its amplitudes) for each one of the micrographs (Figure 19A). Next, these amplitudes are classified following a strategy based on a multivariate statistical analysis (Section 3.1.4), generating class-averages of the amplitude spectra. Each of the generated class-averages, is an average of a certain number of the individual amplitudes of the micrographs. Therefore, the visualization of the Thon rings is.

(47) 43. improved in the class-averages when compared to each one of its members (Figures 19A and 20A).. Figure 19 - Amplitude spectrum and CTF profile. (A) It is shown the amplitude spectrum calculated for one micrograph. (B) The average of all the amplitude spectra calculated for each one of the micrographs. The Thon rings are improved. (C) An amplitude variation profile was created along a central line of the average amplitude spectrum (red central line in B). This profile shows how the CTF is transmitting the information along the spatial frequencies. Source: By the author.. Next, a ring-shaped mask is applied to each of the class-averages (Figure 20B). For this masking process, the internal and external radius of the mask will be selected according to the observed amplitude oscillation profile in a radial direction of the spectrum (Figure 19B, red line, and Figure 19C, profile plot). The selected region, according to the profile plot, is from the first zero-crossing of the contrast transfer function (CTF) until the region where no more oscillations are observed. Inside this region, theoretical CTFs will be generated, based on experimental parameters and estimating the defocus and astigmatism (defocus angle and intensity), for each class-average. Then, the estimated CTFs (Figure 20C, down right-half) are compared to the corresponding input experimental CTF (Figure 20C, up left-half) until the most similar is found. The estimated CTF parameters together with the experimental ones will be used to correct the images for the CTF. In the "half-half" images showed in figure 20a, the equivalence of the Thon rings can be used to check the accuracy of the CTF estimation, where the Thon rings of both.

(48) 44. half should fit. Those class-averages showing the better adjustment are selected. Knowing the individual members (amplitudes) of the classes allows to extract the corresponding micrographs from the raw data and undergo a phase-flipping correction process with their own defocus and astigmatism values.. Figure 20 - (A) A class-average of some individual amplitude spectrum (C) A ring masked classaverage, which will be then used to estimate the CTF, as well as the defocus and astigmatism direction (defocus angle). (D) The estimated CTF (down right-half) will be compared to the experimental or input CTF (up left-half), to then perform the CTF-flip process. Source: By the author.. 3.1.3. Particle Picking. Once the CTF flip is already done, the particle picking can be performed on the "corrected" micrographs. Particles can be selected in many ways. In the ABC methodology, it is suggested to use the local variance to identify and perform the particle picking, avoiding any bias in the process.65 Particle picking can also be done by cross-correlation, using filtered projection images from a known model or, as an initial approach, a few rotationally averaged particles. 64 In the beginning, 3 to 6 different particles of different views are selected from some particular micrograph.

(49) 45. and extracted in square boxes. Then these particles are centered, masked by a circular mask to reduce the influence of the neighboring particles and background, and finally rotationally averaged. In Figure 21, are shown an example of four selected particles (top row), masked and centered, and their corresponding rotational-averages (bottom row).. Figure 21 – Some selected particles (top row) and their corresponding rotational-averages image that will be then used as reference to perform a particle picking by cross-correlation Source: Adapted from AFANASYEV.64. These rotational-average images are used as featureless templates to perform a particle picking by cross-correlation, on all the "corrected" micrographs. After this first round of particle picking, a lot of junk (thick carbon, ice, molecules aggregates, etc.) will be present in the particle data set. So, this undesired particles can be excluded based on histograms distribution of different criteria (e.g.: standard deviation, average density, cross-correlation coefficient, etc.).64. 3.1.4. Multivariate statistical analysis (MSA): Hyperspace data compression and classification. A critical step in image analysis is to organize in the better viable way, all the information in the large particle data set. In this sense, we know that each particle is a 2D image that can be represented by a certain number of pixels being each pixel a specific value of density. Thus, if we consider that each particle is, say an image.

(50) 46. of 100 x 100 pixels in size, then we have 10,000 values of density for each of these particles. Also, if our data set, has about 50,000 particles, all with the same size, then, each of these particles can be represented by a vector with 10,000 values in its array. It means that one would be dealing with a 50,000 x 10,000 matrix of pixels. Alternatively, we can consider each image as a point in a multi-dimensional space, known as hyperspace, where the dimension is the same of the number of pixels in the image.66 Since, in this hyperspace, each particle is represented by a point, those particles which are similar will be separated by shorter distances than those particles that are quite different. Unfortunately, it does not help much on the complexity of the problem since this hyperspace, following the example, would consist of 50,000 particles or points, a data cloud, represented in a 10,000dimensional hyperspace. In Figure 22 A is represented the simplest case of a set of images, with two pixels each one. Therefore, the principal idea of the MSA, is to optimize the representation of the data cloud, through a principal component analysis of the data set, in such a way the new principal axis will coincide with the largest elongation of the cloud or the "greatest variance of the dataset". In Figure 22 B, these new axes are represented by green arrows, and the corresponding vectors (called eigenvectors) are indicated as "u", in the direction of the greatest variance of the dataset, and "v", in direction of the remaining greatest variance. This second direction represents only small modulations with respect to the main trend of the data set and may be ignored in this case. Therefore, in Figure 22, the points of the cloud can be represented in function of the main variance of the cloud using the vector “u”. This is the basic idea of the MSA: the hyperspace data compression. Hence, after performing the MSA, the huge number of coordinates is reduced to just few "principal" components, so the particles can now be represented as a linear combination of some "eigenvectors", which represent the most important variations in the data set, weighting them according to different "eigenvalues".66,67.

(51) 47. Figure 22 –. Multivariate Statistical Analysis (MSA). A) Representation of the simplest case, of images with two pixels each one. Each number inside a pixel, represents a density value. B) Representation of each image in (A), in the hyperspace. Now, each image is represented as a point, according to its density values. The aim of the MSA, is to optimize the representation of the data cloud, through a rotation and translation of the coordinate system. Therefore, the new principal axis will coincide with "greatest variance of all dataset". These new axes are represented by green arrows, and the corresponding eigenvectors are presented as "u", in the direction of the greatest variance of all dataset, and "v", in direction of the remaining greatest variance. This last direction represents only small modulations with respect to the main trend of the data set and may be ignored. Therefore, the points of the cloud can be represented only in function of the main variance of the cloud, thus reducing the number of coordinate necessaries to define and image in the hyperspace. Source: Adapted from VAN HEEL.67. After performing the MSA, the images, or the points in the hyperspace, which are sharing common features (for example, showing the particle in similar orientations) are grouped into "sub-clouds" or "classes" by an intra-class variance minimization, which is equivalent to the maximization of the inter-class variance. Once determined which particles belong to each class, an averaged of those particles will be performed, to result in a class-average, which will have a better signal to noise ratio (SNR) than their individual members. Commonly the number of classes-averages is chosen to have around few tens of particles per class, however, it will vary according to the quantity and quality of the data set.. 3.1.5. Euler angular assignment The class-averages or “classums” must be subjected to a process of. assigning Euler angles to allow them to be available to the 3D reconstruction process. There are two methods of assigning Euler angles: angular reconstitution and projection.

(52) 48. matching. The former is an unsupervised method, since it finds the intrinsic angular orientation between images based on their own information. The latter is a supervised method that uses projections of a given 3D model, generated in different spatial directions. Then, comparing these projections to the classaverages, the Euler angles can be assigned. This method when performed without the necessary caution may result in some bias induced by the model. 54 In the current work, it will be used the angular reconstitution method, which is based on the “common line projection theorem". The underlying idea of this theorem is that, if there are two 2D-projections from a 3D volume, these share at least one 1D-projection, or a “common line”. Therefore, if one could idetify the common line between each pairwise of classes (from all dataset), then the Euler angles could be assigned for each class-average.68,69 In Figure 23, it is shown a graphic representation of how the angularreconstitution process works. In the beginning, each 2D-class is projected into a 1D-line. Then, the class is rotated one degree (1°) in the plane, and again projected into a 1D-line, and so on over a 180° range. Therefore, all this set of 1Dprojections from one specific class can be displayed as a single 2D-image known as a “sinogram” of the corresponding class (Figure 23: S1, S2, and S3). All sinograms are then compared pairwise in the following manner: the crosscorrelation coefficient between each line of the first and each line of the second sinogram are computed. All correlation coefficients are then stored into a sinogram correlation function (SCF), that is represented as a 2D-image (CSC12 and CSC32). After to apply a fitting procedure over this SCF, the "common line" or "common tilt axis" between the two projections can be found (blue lines, common lines between projections 2 and 1, while red lines indicate the common lines between projections 2 and 3).68,69 Also, the angular distance between the common lines (blue and red) gives the angle between projections 1 and 3. Hence, starting with a set of three projection images, it is possible to find the common tilt axis between each given pair of projections..

(53) 49. Figure 23 - Sinograms and sinogram correlation function for a model structure. There are shown three projections (#1, #2, #3), from a model composed of three Gaussian dots with different densities. The 1D projections set or sinograms are namely S1, S2 and S3, for each 2D projection respectively. CSC12 and CSC32 represent cross-sinogram correlation functions between projections 1 and 2 and projections 3 and 2, respectively. Each point of the sinogram correlation function contains the correlation coefficient of a pair of lines from the two sinograms. Blue solid lines indicate the common lines between projections 2 and 1, while the red solid lines indicate the common lines between projections 2 and 3. Each CSC has two peaks because projections from 180° to 360° mirror those from 0° to 180°. The angular distance between the common lines (blue and red) gives the angle between projections 1 and 3. Source: Adapted from ORLOVA.70. 3.1.6. Three-dimensional reconstruction. Once the Euler angles were already assigned to the different class-averages or "classums", they can be used for the three-dimensional reconstruction process. To carry out this task, the class-averages are back-projected into a 3D-space.48,68 In Figure 24, it is shown a schematic representation of how the three-dimensional reconstruction is performed from the 2D-classes. After performing a 3Dreconstruction, this 3D map is re-projected into the same directions as the classaverages in a way that each class can be compared with its respective reprojection, and the similarity degree would give a measure of how good the angular assignment was performed..

(54) 50. Figure 24 - The back-projection of the 2D class-average to produce a 3D density map. This 3D density map captures the electron density throughout the macromolecular complex. Source: Adapted from MIT.71. The initial 3D map can be projected into different spatial directions to obtain a new set of projections, also known as "anchor-set", to be used as references, since they already have known angles, to refine the Euler angles assignment of the classes-averages. This is an example of 2D-projections set (“anchor-set”), that could be used to refine the previous angles of the classes-average. This process is iteratively performed until achieving a stable 3D map. All these steps will be detailed in section 5. Another important aspect to mention is that the one can create a separate set of 2D-projections and use them as references to align the original particles. This process known as multi-reference-alignment (MRA),48 works as follows: firstly, each particle will be compared by cross-correlation with each one of all projections. Then, the particles will be aligned (by displacements and rotation in the image plane) with respect to those projections they had shown to have the highest crosscorrelation coefficient (CCC). After these process, the particles can be subjected for a new MSA process and re-classification.64. 3.2. DNA production. A puC-lambda plasmid was used to produce, by Polymerase Chain reaction (PCR), the DNA fragment of 186 bp length, containing the wild-type (wt) λ PR promoter (-100 to +30). Below is the Promoter sequence as well as a representation of the consensus regions..

(55) 51. Figure 25 - Complete sequence of the DNA wild type Lambda PR promoter (in black letters and underlined). Sketch of the DNA promoter in which are remarked the principal regions of the DNA that interacts with the RNAP during the RP complex formation. Source: By the author.. Figure 25 shows the complete sequence of DNA wt λPR promoter (-100 to +30), in black letters underlined, and with the transcription start site in purple letter. The colored shadows are explained in the promoter sketch (bottom). In this sketch, a purple arrow represent the transcription start site (TSS) and also by a yellow rectangle are represented the -10 hexamer region (from -7 to -12) and the -35 hexamer region (from -30 to -35). The discriminator region (from -1 to -6) is represented by an orange rectangle and the spacer zone (from -13 to -29) is represented by a red rectangle. In addition, the AT-rich regions, known as UPelements, are represented by green rectangles: first UP-element (from -45 to -58), the second UP-element (from -68 to – 79) and the third UP-element (from -90 to 100). Once the DNA was amplified by PCR, the purification procedure started with Polyacrylamide Gel Electrophoresis (PAGE), which separates DNA by their base pairs length. After identifying the desired DNA band, it was purified by GeneJET PCR purification Kit (Thermo Fisher Scientific, USA), and its concentration was quantified.. 3.3. Transcription Initiation complex assembly. To assembly the RP complex, the amplified DNA and the E. coli RNAP (New England Biolabs, USA) were used. The complex was prepared in transcription buffer TB5X (Tris-HCl 100 mM, pH 8.0, MgCl2 25 mM, KCl 250 mM), maintaining the RNAP:.

(56) 52. DNA in a 1:2 molar ratio, and according to the protocol described in table IV. The protocol described here, which includes the corresponding binding test (Electrophoretic Mobility Shifting Assays), was widely reported in different previous studies of the same RNAP-DNA complex.22,23,72,73. Table 1 - Protocol for Transcription initiation complex assembly. Source: By the author. (a) Dithiothreitol (b) Once mixed, the mixture is incubated at 37°C for 20 minutes. After this, 2 μL of sodium heparin (2mg/mL) is added, and the mixture is incubated for 20 minutes at 25°C.. 3.4. Specimen preparation and TEM imaging. An ultra-thin carbon coated grid (PELCO, USA) was used for this work. This grid was negatively charged (at 15 mA and 25 seconds) using an easiGlowTM Glow Discharge Cleaning System (PELCO, USA), for a better deposition of the sample on the grid. After that, 3 uL of the RP complex sample (with a concentration around ~ 0.02 mg/mL) were deposited on the grid, and after one minute, the excess of solution was removed using a filter paper. Immediately after, 3uL of uranyl acetate (2% m/v) was applied on the grid for the negatively stain the sample, and after 30 seconds the excess of solution was removed by a filter paper. The uranyl acetate solution was applied twice. Later, the micrographs were collected using the JEM 2100 microscope, (JEOL, Japan) equipped with an F-416 CMOS camera (TVIPS, Germany). Finally, 500 micrographs were collected using 200 kV acceleration.

(57) 53. voltage, at 60,000x of magnification with a pixel size of 1.78 Å and in a constant defocus level of about -1.7 μm.. 3.5. Single particle analysis. All the image processing was carried out using the Imagic4D software (Image Science, Germany) according to the most recent methodology reported in the literature for this package, the Align by Classification process.64 It includes the various stages described in Figure 26, that were previously detailed in section 3.1. This began with the pre-treatment of the raw micrographs, to remove the extreme low and high spatial frequencies and it was followed by the CTF correction by phase-flipping. After that, a first particle picking was performed. These particles were boxed in 180x180 pixels images, then centered and circularly masked. These particles were then normalized and classified into ~2,500 class-averages using a multivariate statistical analysis (MSA) and a hierarchical ascendant classification (HAC).67 The class-averages with extreme lower members per class or with low quality due to thick carbon, grid borders or overlapped particles, were removed and a better particle stack was obtained and re-classified. Once a good set of classaverages was obtained, the particles were centered and circular masked for the angular assignment and posteriorly three-dimensional reconstruction. This first 3Dmap was used to refine the previous steps in the process until obtaining a consistent 3D final map. The resolution of the final 3D-map was calculated using Fourier-Shell-Correlation and the ½-bit criteria.74,75.

(58) 54. Figure 26 - Single particle analysis Cryo-EM workflow using IMAGIC 4D software Source: Adapted from IMAGIC-MANUALS.76. 3.6. Modelling. For the molecular fitting of the RNAP into the 3D density map, different crystallographic structures were used. The RNAP holoenzyme and the DNA transcription bubble were fitted using the available crystal structure of the E. coli Transcription Initiation Complexes with a complete bubble, considering a promoter with 17 bp spacer.6 The DNA sequence of the complete transcription bubble, until the -35 hexamer region, was replaced by the sequence of the promoter used in this study. The DNA region upstream the -35 hexamer toward -90, was modeled by segments using 3D-DART.77 Then the DNA segments were manually fitted into the 3D map, following the strong observable density in it, and using Chimera package. The interaction of the αCTDs with the UP-element regions of the promoter was fitted according to the available structures of this domain in complex with CAP and a DNA region.18 Finally, the N-terminal σ70 subunit, the domain 1.1, was taken from the structure reported by Bae et al. and manually fitted.13 For all the rigidbody and manual fitting processes, it was used the UCSF Chimera 1.12 software.78.